MCP Server 高级配置教程

MCP 协议简介

Model Context Protocol (MCP) 是一个开放协议，用于连接 AI 模型与外部工具和数据源。OpenClaw 作为 MCP Client，可以连接多个 MCP Server，让 AI 助手获得访问数据库、文件系统、API 等外部资源的能力。

查看 MCP 连接

openclaw mcp list

MCP Servers:
  Name            Transport   Status    Tools
  ──────────────────────────────────────────────
  filesystem      stdio       ● Online  5
  database        sse         ● Online  3
  github          stdio       ● Online  8
  slack-tools     stdio       ○ Offline 0

添加 MCP Server

Stdio 传输方式

{
  "mcp": {
    "servers": {
      "filesystem": {
        "transport": "stdio",
        "command": "npx",
        "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed/dir"],
        "env": {}
      }
    }
  }
}

SSE 传输方式

{
  "mcp": {
    "servers": {
      "remote-tools": {
        "transport": "sse",
        "url": "https://mcp-server.example.com/sse",
        "headers": {
          "Authorization": "Bearer {{MCP_SERVER_TOKEN}}"
        }
      }
    }
  }
}

常用 MCP Server 配置

文件系统

{
  "mcp": {
    "servers": {
      "filesystem": {
        "transport": "stdio",
        "command": "npx",
        "args": ["-y", "@modelcontextprotocol/server-filesystem", "/home/user/documents"]
      }
    }
  }
}

PostgreSQL 数据库

{
  "mcp": {
    "servers": {
      "postgres": {
        "transport": "stdio",
        "command": "npx",
        "args": ["-y", "@modelcontextprotocol/server-postgres"],
        "env": {
          "POSTGRES_URL": "{{DATABASE_URL}}"
        }
      }
    }
  }
}

GitHub

{
  "mcp": {
    "servers": {
      "github": {
        "transport": "stdio",
        "command": "npx",
        "args": ["-y", "@modelcontextprotocol/server-github"],
        "env": {
          "GITHUB_TOKEN": "{{GITHUB_TOKEN}}"
        }
      }
    }
  }
}

Web 搜索

{
  "mcp": {
    "servers": {
      "brave-search": {
        "transport": "stdio",
        "command": "npx",
        "args": ["-y", "@modelcontextprotocol/server-brave-search"],
        "env": {
          "BRAVE_API_KEY": "{{BRAVE_API_KEY}}"
        }
      }
    }
  }
}

为频道分配 MCP 工具

{
  "channels": {
    "telegram-main": {
      "mcpServers": ["filesystem", "brave-search"],
      "tools": ["mcp:filesystem:read_file", "mcp:brave-search:search"]
    }
  }
}

查看 MCP Server 提供的工具

openclaw mcp tools filesystem

Tools from 'filesystem':
  read_file        Read contents of a file
  write_file       Write content to a file
  list_directory   List files in a directory
  search_files     Search for files by name
  get_file_info    Get file metadata

MCP Server 健康检查

openclaw mcp health

MCP Server Health:
  filesystem     ✓ Healthy   Tools: 5   Latency: 5ms
  database       ✓ Healthy   Tools: 3   Latency: 15ms
  github         ✓ Healthy   Tools: 8   Latency: 120ms
  slack-tools    ✗ Offline   Connection refused

安全配置

限制 MCP 工具的使用权限：

{
  "mcp": {
    "security": {
      "requireApproval": ["write_file", "delete_file", "execute_query"],
      "blockedTools": ["drop_table"],
      "maxCallsPerMinute": 30,
      "timeout": 10000
    }
  }
}

requireApproval 列表中的工具在执行前会请求用户确认。

开发自定义 MCP Server

创建一个简单的 MCP Server：

import { Server } from "@modelcontextprotocol/sdk/server/index.js";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";

const server = new Server({
  name: "my-tools",
  version: "1.0.0"
}, {
  capabilities: { tools: {} }
});

server.setRequestHandler("tools/list", async () => ({
  tools: [{
    name: "get_time",
    description: "获取当前时间",
    inputSchema: { type: "object", properties: {} }
  }]
}));

server.setRequestHandler("tools/call", async (request) => {
  if (request.params.name === "get_time") {
    return {
      content: [{ type: "text", text: new Date().toISOString() }]
    };
  }
});

const transport = new StdioServerTransport();
await server.connect(transport);

在 OpenClaw 中注册：

{
  "mcp": {
    "servers": {
      "my-tools": {
        "transport": "stdio",
        "command": "node",
        "args": ["./my-mcp-server.js"]
      }
    }
  }
}

调试 MCP 连接

openclaw mcp test filesystem
openclaw mcp logs filesystem --last 20

总结

MCP 协议为 OpenClaw 提供了标准化的工具扩展机制。通过连接各种 MCP Server，AI 助手可以访问文件系统、数据库、外部 API 等丰富的外部资源，真正实现「AI + 工具」的强大组合。